The scientific origins of the Dictator/Sitcom Character game

There’s a game where your computer will attempt to guess what dictator or sitcom character you’re thinking of. It’s based on a binary tree, where a series of questions about whether you’re thinking of this or that narrow down the number of possible solutions. You can use it to guess the names of dictators, but the same technique has been used to identify many other items. Using binary trees to make a dichotomous key can help you think about anything you want to classify. So for plants you could start by asking if it’s a flowering or non-flowering plant. If it is flowering does it produce one seed or many? This continues until you have a plant that is this and not that. It’s thought that Lamarck published the first dichotomous key for plants in 1778, but the Royal Society’s website suggests that Lamarck was building on work from a century earlier.

Obviously Lamarck didn’t consult the Royal Society’s website. Instead the Royal Society have digitised British grasses and wildflowers by Richard Waller as part of their Turning the Pages project. Lawrence Griffing has had a close look at this in the paper Who Invented the Dichotomous Key? Richard Waller’s Watercolors of the Herbs of Britain in the American Journal of Botany. He’s found there’s more to the images than simply their looks.

There’s a question of how you choose what kind of branches to make. You could start by asking “Is this something you find in Larry’s garden, yes/no?”. It’s eccentric, but with enough questions you could eventually identify a plant. With evolution it becomes far more sensible to ask questions based on common descent, but Waller was working before there was a well-accepted theory of natural selection. What Waller could do though is observe the structure of plants closely. This is what he did. Griffing says what Waller was doing is something called a “linear leaf-ordering of hierarchical clusters”. This is a technique used today in Bioinformatics (and you can pick up free access to a paper on it from the journal Bioinformatics).

I’m wary of saying that a current technique is identical to Waller’s work from the early modern period. Further, Griffing’s paper shows that Waller’s work was not taken up by the Royal Society, so there’s no direct line between Waller’s clustering and modern clustering. However, the fact that Waller continued his work shows an intense interest in plant classification. Even if the Royal Society weren’t interested, it’s likely that Waller did have colleagues and acquaintances who had an interest in his work and methods.

There used to be a very Whiggish tendency in science history that what we know now is an inevitable consequence of what we knew then. If you think that then it becomes easy to pull out a single narrative tracking the single path to our current state. What Griffing’s paper shows is that it’s not always that neat a narrative.

References

Bar-Joseph Z., Gifford D.K. & Jaakkola T.S. (2001). Fast optimal leaf ordering for hierarchical clustering, Bioinformatics, 17 (Suppl 1) S22-S29. DOI: 10.1093/bioinformatics/17.suppl_1.S22

Griffing L.R. (2011). Who invented the dichotomous key? Richard Waller’s watercolors of the herbs of Britain, American Journal of Botany, 98 (12) 1911-1923. DOI: 10.3732/ajb.1100188

Images

Illustration from ‘British grasses and wildflowers by Richard Waller’ hosted by the Royal Society.